In an existing communications system, for example, in a universal mobile telecommunications system (hereinafter referred to as UMTS), speech encoding adopts a large amount of convolutional codes as channel encoding and utilizes a power control mechanism to guarantee its speech quality. FIG. 1 is a schematic diagram of a system architecture of a speech encoding process in the prior art. As shown in FIG. 1, taking a UMTS network as an example, a processing process of an uplink adaptive multi-rate (hereinafter referred to as AMR) speech signal is that, speech encoding of an AMR speech encoder (hereinafter referred to as AMR Speech Encoder) in a user equipment (hereinafter referred to as UE) adopts convolutional codes through a convolutional code encoder (hereinafter referred to as CC Encoder) to perform encoding processing, and AMR speech signals after the encoding of the CC Encoder are sent to a base station (hereinafter referred to as NodeB) through an air interface; a CC decoder (hereinafter referred to as CC Decoder) in the NodeB can decode the AMR speech signals, and the CC Decoder includes two outputs. In one output, a decoded bit stream is sent to a radio network controller (hereinafter referred to as RNC) through an Iub interface, and then is sent by the RNC to an AMR speech decoder (hereinafter referred to as AMR Speech Decoder) in a core network (hereinafter referred to as CN) through an Iu interface. In the other input, a cyclic redundancy check (hereinafter referred to as CRC) check result, that is, a CRC indicator (hereinafter referred to as CRCI), is sent to the RNC through the Iub interface, and the RNC then can send a bad frame indicator (hereinafter referred to as BFI) to the AMR Speech Decoder in the CN through the Iu interface according to the CRCI. The CC Decoder also sends the CRCI to an outer-loop power control module (hereinafter referred to as Outer-Loop Power Control) in the RNC. After receiving the decoded bit stream and the BFI, the AMR Speech Decoder can perform decoding processing; while the Outer-Loop Power Control may adjust a target block error ratio (hereinafter referred to as BLER) according to the CRCI, and send a target signal-to-interference-plus-noise ratio (hereinafter referred to as Target SINR) to an inner-loop power control module (hereinafter referred to as Inner-Loop Power Control) in the NodeB according to an adjusted BLER. The Inner-Loop Power Control sends a power command (hereinafter referred to as Power Command) to a power transmitter (hereinafter referred to as Power Transmitter) of the UE, according to a measured signal-to-interference-plus-noise ratio (hereinafter referred to as Measured SINR) and the Target SINR, to adjust transmission power of the UE. FIG. 2 is a schematic structural diagram of processing three substreams in the system architecture which is shown in FIG. 1. As shown in FIG. 2, in the prior art, the AMR speech signal may be classified into three substreams, A, B, and C, that is, Class A signal, Class B signal, and Class C signal. Substream A has the strongest influence on the speech quality and is also the most important, after whose data block, a 12-bit CRC is attached. Substreams B and C are relatively less important and the data block is followed with no CRC. The CC Decoder in the NodeB adopts a Viterbi algorithm (hereinafter referred to as VA) decoder, while in a decoding result of the VA decoder, only substream A has the CRCI.
However, during an implementation process of the present invention, the inventor finds that in the prior art the NodeB has a relatively low decoding performance for convolutional codes of substream A, which has a relatively strong influence on the speech quality and fails to meet users' higher requirements for the speech quality.